Google-Cloud-Platform
我對私有 GKE 對等互連和 Cloud NAT 做錯了什麼？
我正在嘗試在 Google Cloud 上安裝以下平台：
2 個私有（原生）GKE 集群，位於 2 個不同的 VPC 中，為了讓它們能夠訪問網際網路，每個 vpc 都配置了一個 Cloud Nat。
我需要的是 2 個 GKE 集群互動，但對等 VPC，我只能在 POD 之間進行通信，而不是在 POD -> 服務或 POD -> 內部負載均衡器之間進行通信。
集群：
NAME LOCATION MASTER_VERSION MASTER_IP MACHINE_TYPE NODE_VERSION NUM_NODES STATUS Shrek01 asia-east1-a 1.16.8-gke.15 <none> g1-small 1.16.8-gke.15 3 RUNNING Shrek02 asia-east2-a 1.15.9-gke.24 <none> g1-small 1.15.9-gke.24 3 RUNNING虛擬個人電腦：
NAME SUBNET_MODE BGP_ROUTING_MODE IPV4_RANGE GATEWAY_IPV4 Shrek01 CUSTOM REGIONAL Shrek02 CUSTOM REGIONAL子網：
NAME REGION NETWORK RANGE Shrek01 asia-east1 Shrek01 192.168.13.0/24 Shrek02 asia-east2 Shrek02 192.168.14.0/24對等：
NAME NETWORK PEER_PROJECT PEER_NETWORK AUTO_CREATE_ROUTES STATE STATE_DETAILS Shrek01-Shrek01-peering Shrek01 pocprod2-2019001 Shrek02 True ACTIVE [2020-05-16T14:29:57.864-07:00]: Connected. Shrek02-Shrek01-peering Shrek02 pocprod2-2019001 Shrek01 True ACTIVE [2020-05-16T14:29:57.864-07:00]: Connected.防火牆規則：
- “Shrek01-對等入口”
{ "allowed": [ { "IPProtocol": "all" } ], "creationTimestamp": "2020-05-16T16:05:14.829-07:00", "description": "", "direction": "INGRESS", "disabled": false, "id": "6807007164648771397", "kind": "compute#firewall", "logConfig": { "enable": false }, "name": "peering-ingress", "network": "https://www.googleapis.com/compute/v1/projects/pocprod2-2019001/global/networks/Shrek01", "priority": 1000, "selfLink": "https://www.googleapis.com/compute/v1/projects/pocprod2-2019001/global/firewalls/peering-ingress", "sourceRanges": [ "192.168.14.0/24", "10.113.64.0/19", "10.213.64.0/19" ] }
- “Shrek02-對等入口”
{ "allowed": [ { "IPProtocol": "all" } ], "creationTimestamp": "2020-05-16T16:24:28.545-07:00", "description": "", "direction": "INGRESS", "disabled": false, "id": "7130188648920500419", "kind": "compute#firewall", "logConfig": { "enable": false }, "name": "Shrek02-peering-ingress", "network": "https://www.googleapis.com/compute/v1/projects/pocprod2-2019001/global/networks/Shrek02", "priority": 1000, "selfLink": "https://www.googleapis.com/compute/v1/projects/pocprod2-2019001/global/firewalls/Shrek02-peering-ingress", "sourceRanges": [ "192.168.13.0/24", "10.113.32.0/19", "10.213.32.0/19" ] }k8s Shrek01集群：
kubectl get svc -o wideNAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE SELECTOR kubernetes ClusterIP 10.213.32.1 <none> 443/TCP 85m <none> nginx LoadBalancer 10.213.60.14 192.168.13.7 80:32612/TCP 92s app=nginx nginx-cip ClusterIP 10.213.34.24 <none> 80/TCP 93s app=nginx nginx-np NodePort 10.213.35.31 <none> 80:30444/TCP 92s app=nginx
kubectl get pod -o wideNAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES nginx-64b4f9bb85-9sjcp 1/1 Running 0 3m34s 10.113.34.11 gke-Shrek01-default-pool-f9ecbfcc-dz9z <none> <none> nginx-64b4f9bb85-l2bzd 1/1 Running 0 3m34s 10.113.32.5 gke-Shrek01-default-pool-f9ecbfcc-pdll <none> <none> nginx-64b4f9bb85-xd7kw 1/1 Running 0 3m34s 10.113.33.9 gke-Shrek01-default-pool-f9ecbfcc-v67d <none> <none>
kubectl get nodes -o wideNAME STATUS ROLES AGE VERSION INTERNAL-IP EXTERNAL-IP OS-IMAGE KERNEL-VERSION CONTAINER-RUNTIME gke-Shrek01-default-pool-f9ecbfcc-dz9z Ready <none> 89m v1.16.8-gke.15 192.168.13.4 Container-Optimized OS from Google 4.19.109+ docker://19.3.1 gke-Shrek01-default-pool-f9ecbfcc-pdll Ready <none> 89m v1.16.8-gke.15 192.168.13.2 Container-Optimized OS from Google 4.19.109+ docker://19.3.1 gke-Shrek01-default-pool-f9ecbfcc-v67d Ready <none> 89m v1.16.8-gke.15 192.168.13.3 Container-Optimized OS from Google 4.19.109+ docker://19.3.1
- 從 Shrek02 的 pod 檢查：
root@nginx-5c66c56f55-8jwv2:/# echo ${MY_POD_IP} 10.113.66.9 # internal load balancer root@nginx-5c66c56f55-8jwv2:/# nc -vz 192.168.13.7 80 192.168.13.7: inverse host lookup failed: Unknown host (UNKNOWN) [192.168.13.7] 80 (?) : Connection timed out # intarnal load balancer's Cluster IP root@nginx-5c66c56f55-8jwv2:/# nc -vz 10.213.60.14 80 10.213.60.14: inverse host lookup failed: Unknown host (UNKNOWN) [10.213.60.14] 80 (?) : Connection timed out # ClusterIP root@nginx-5c66c56f55-8jwv2:/# nc -vz 10.213.34.24 80 10.213.34.24: inverse host lookup failed: Unknown host (UNKNOWN) [10.213.34.24] 80 (?) : Connection timed out # NodePort root@nginx-5c66c56f55-8jwv2:/# nc -vz 10.213.35.31 80 10.213.35.31: inverse host lookup failed: Unknown host (UNKNOWN) [10.213.35.31] 80 (?) : Connection timed out # Pod IP root@nginx-5c66c56f55-8jwv2:/# nc -vz 10.113.34.11 80 10.113.34.11: inverse host lookup failed: Unknown host (UNKNOWN) [10.113.34.11] 80 (?) open root@nginx-5c66c56f55-8jwv2:/# nc -vz 10.113.32.5 80 10.113.32.5: inverse host lookup failed: Unknown host (UNKNOWN) [10.113.32.5] 80 (?) open root@nginx-5c66c56f55-8jwv2:/# nc -vz 10.113.33.9 80 10.113.33.9: inverse host lookup failed: Unknown host (UNKNOWN) [10.113.33.9] 80 (?) open我忘了任何步驟嗎？我沒發現錯誤。
- 我能夠將 pod 連接到 pod，將 pod 連接到節點埠，並將埠連接到具有類似環境的外部負載均衡器，這是警告：
只有在以下情況下，VPC 對等互連才允許使用內部負載均衡器：
- 對等網路中的客戶端虛擬機 (VM) 實例與您的內部負載均衡器位於同一區域
- 您配置全域訪問。配置全域訪問後，來自對等 VPC 網路任何區域的客戶端虛擬機實例都可以訪問您的內部 TCP/UDP 負載均衡器。內部 HTTP(S) 負載平衡不支持全域訪問。
為了將內部全域訪問與來自不同區域的 VPC 對等連接使用，您有兩種選擇：
- 如果您知道內部負載均衡器名稱，則可以使用以下命令：
$ gcloud compute forwarding-rules update <LB_NAME> \ --region=<REGION> \ --allow-global-access
- 並檢查它：
gcloud compute forwarding-rules describe <LB_NAME> \ --region=us-west1 \ --format="get(name,region,allowGlobalAccess)"
如果您所在地區的內部 LB 很少，其他簡單的更改方法是通過 GCP 負載平衡頁面點擊此處訪問：
- 在Name列中，點擊集群區域的內部 TCP/UDP 負載均衡器（點擊它後，您將看到子網名稱，如下面的範例所示）：
然後點擊編輯。
點擊前端配置
點擊鉛筆進行編輯
在全域訪問下，選擇 啟用。
點擊完成
打開並點擊更新以更新規則。
- 等到規則完成應用。
- 更改後，我能夠從 Shrek02 定位到 Shrek01 中的內部 LB。（我將在下面給出範例）。
注意：：
ClusterIP在集群內部 IP 上公開服務。選擇此值使服務只能從集群內訪問。所以你不會被路由到外部訪問。再生產：
- 我按照您的參數創建了兩個 VPC：
$ gcloud container clusters list NAME LOCATION MASTER_VERSION MASTER_IP MACHINE_TYPE NODE_VERSION NUM_NODES STATUS shrek01 europe-west1-b 1.16.8-gke.15 XX.XXX.XX.XXX g1-small 1.16.8-gke.15 3 RUNNING shrek02 europe-west2-b 1.15.9-gke.24 XXX.XXX.XX.XXX g1-small 1.15.9-gke.24 3 RUNNING $ gcloud compute networks subnets list NAME REGION NETWORK RANGE shrek01 europe-west1 shrek01 192.168.13.0/24 shrek02 europe-west2 shrek02 192.168.14.0/24 $ gcloud compute networks peerings list-routes sh1-sh2 --network=shrek01 --region europe-west1 --direction=INCOMING DEST_RANGE TYPE NEXT_HOP_REGION PRIORITY STATUS 192.168.14.0/24 SUBNET_PEERING_ROUTE europe-west2 1000 accepted 10.229.0.0/20 SUBNET_PEERING_ROUTE europe-west2 1000 accepted 10.36.0.0/14 SUBNET_PEERING_ROUTE europe-west2 1000 accepted $ gcloud compute networks peerings list-routes sh2-sh1 --network=shrek02 --region europe-west2 --direction=INCOMING DEST_RANGE TYPE NEXT_HOP_REGION PRIORITY STATUS 192.168.13.0/24 SUBNET_PEERING_ROUTE europe-west1 1000 accepted 10.154.0.0/20 SUBNET_PEERING_ROUTE europe-west1 1000 accepted 10.24.0.0/14 SUBNET_PEERING_ROUTE europe-west1 1000 accepted在確保我的節點可以在 VPC 之間 Ping 之後，我將使用這個 yaml 測試入口和連接
hello-1.yaml:apiVersion: apps/v1 kind: Deployment metadata: name: hello-1 spec: replicas: 3 selector: matchLabels: app: hello-1 template: metadata: labels: app: hello-1 spec: containers: - name: hello-1 image: gcr.io/google-samples/hello-app:1.0 ports: - name: http containerPort: 8080 --- apiVersion: v1 kind: Service metadata: name: hello-1-svc spec: type: NodePort selector: app: hello-1 ports: - protocol: TCP port: 80 targetPort: 8080
hello-2.yaml:apiVersion: apps/v1 kind: Deployment metadata: name: hello-2 spec: replicas: 3 selector: matchLabels: app: hello-2 template: metadata: labels: app: hello-2 spec: containers: - name: hello-2 image: gcr.io/google-samples/hello-app:2.0 ports: - name: http containerPort: 8080 --- apiVersion: v1 kind: Service metadata: name: hello-2-svc spec: type: NodePort selector: app: hello-2 ports: - protocol: TCP port: 80 targetPort: 8080
hello-ingress.yaml:apiVersion: networking.k8s.io/v1beta1 kind: Ingress metadata: name: hello-ingress annotations: kubernetes.io/ingress.class: nginx spec: rules: - host: http: paths: - path: / backend: serviceName: hello-1-svc servicePort: 80 - path: /v2 backend: serviceName: hello-2-svc servicePort: 80
- 查看 pod 名稱和 pod 的 IP，以及節點 IP 和負載均衡器/節點埠埠。
o$ kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES hello-1-84d5994678-dx8dv 1/1 Running 0 140m 10.24.0.9 gke-shrek01-default-pool-5ffc38d7-bz35 <none> <none> hello-1-84d5994678-t74mn 1/1 Running 0 14m 10.24.1.3 gke-shrek01-default-pool-5ffc38d7-70sk <none> <none> hello-1-84d5994678-zq7t2 1/1 Running 0 14m 10.24.2.9 gke-shrek01-default-pool-5ffc38d7-zfj6 <none> <none> hello-2-5c4f554ccc-b8j6f 1/1 Running 0 140m 10.24.0.10 gke-shrek01-default-pool-5ffc38d7-bz35 <none> <none> hello-2-5c4f554ccc-km4ph 1/1 Running 0 13m 10.24.1.4 gke-shrek01-default-pool-5ffc38d7-70sk <none> <none> hello-2-5c4f554ccc-z4f6n 1/1 Running 0 13m 10.24.2.10 gke-shrek01-default-pool-5ffc38d7-zfj6 <none> <none> $ ubectl get services NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE hello-1-svc NodePort 10.154.13.186 <none> 80:32186/TCP 140m hello-2-svc NodePort 10.154.4.214 <none> 80:32450/TCP 140m $ kubectl get svc ingress-nginx-controller -n ingress-nginx NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE ingress-nginx-controller LoadBalancer 10.154.10.104 192.168.13.5 80:30112/TCP,443:32156/TCP 4h20m $ kubectl get ingress NAME HOSTS ADDRESS PORTS AGE hello-ingress * 192.168.13.5 80 98m $ kubectl get nodes -o wide NAME STATUS ROLES AGE VERSION INTERNAL-IP EXTERNAL-IP OS-IMAGE KERNEL-VERSION CONTAINER-RUNTIME gke-shrek01-default-pool-5ffc38d7-70sk Ready <none> 2d19h v1.16.8-gke.15 192.168.13.3 XX.XXX.XX.XXX Container-Optimized OS from Google 4.19.109+ docker://19.3.1 gke-shrek01-default-pool-5ffc38d7-bz35 Ready <none> 2d19h v1.16.8-gke.15 192.168.13.2 XXX.XXX.XX.XXX Container-Optimized OS from Google 4.19.109+ docker://19.3.1 gke-shrek01-default-pool-5ffc38d7-zfj6 Ready <none> 2d19h v1.16.8-gke.15 192.168.13.4 XX.XXX.X.XXX Container-Optimized OS from Google 4.19.109+ docker://19.3.1現在我將連接到
shrek02集群，創建一個 pod 並安裝curl：project@cloudshell:~$ kubectl run ubuntu --image=ubuntu -it -- /bin/bash root@ubuntu:/# apt update root@ubuntu:/# apt install curl root@ubuntu:/# exit project@cloudshell:~$ kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES ubuntu 1/1 Running 1 2m51s 10.36.1.6 gke-shrek02-default-pool-a7a08ac8-0lrz <none> <none>
- 您可以看到我們在 中
shrek02，現在讓我們測試與shrek01資源的連接。請記住，這kube-dns僅在集群內部可用，因此我們將使用 IP 進行連接：project@cloudshell:~$ kubectl exec -it ubuntu -- /bin/bash ###Hello-1 POD: root@ubuntu:/# curl 10.24.0.9:8080 Hello, world! Version: 1.0.0 Hostname: hello-1-84d5994678-dx8dv ###Hello-2 POD: root@ubuntu:/# curl 10.24.1.4:8080 Hello, world! Version: 2.0.0 Hostname: hello-2-5c4f554ccc-km4ph ### HELLO-1-SVC USING NODE IP + NODEPORT: root@ubuntu:/# curl 192.168.13.3:32186 Hello, world! Version: 1.0.0 Hostname: hello-1-84d5994678-t74mn ### HELLO-2-SVC USING ANOTHER NODE IP + NODEPORT: root@ubuntu:/# curl 192.168.13.2:32450 Hello, world! Version: 2.0.0 Hostname: hello-2-5c4f554ccc-km4ph ### NOW LET'S TEST OUR INGRESS which routes "/" to hello-1 and "/v2" to hello-2: root@ubuntu:/# curl 192.168.13.5/ Hello, world! Version: 1.0.0 Hostname: hello-1-84d5994678-dx8dv root@ubuntu:/# curl 192.168.13.5/v2 Hello, world! Version: 2.0.0 Hostname: hello-2-5c4f554ccc-b8j6f我希望它可以幫助您解決環境問題，如果您有任何問題，請在評論中告訴我。
